# Plan: Aurelian Frontier Followups

## Overview

Iterate the Aurelian Frontier proof slice -- a generated-content adventure
demo whose authoritative scenario process owns regional markets, calendar
events, construction jobs, party transitions, agent NPC budgets, and
authority-combat verbs. Long-form decomposition lives in
`docs/backlog/aurelian-frontier.md`; the proof slice description is in
`docs/demos/adventure.md` and the deterministic mission slice is captured
in commit history (see `WORKPLAN.md` observable milestones for hash-pinned
landings).

Goal of each iteration: pull one or two open follow-ups from
`docs/backlog/aurelian-frontier.md` into reviewable shape, exercise them
through the real scenario process under `make run-adventure`, and keep
generated content in `demos/adventure-content/content/prototype.cue`
plus `demos/adventure-content/` deterministic.

## Conflict Surface

Owned by this plan:

- `demos/adventure-content/` (including
  `demos/adventure-content/content/prototype.cue`),
  `demos/adventure-server/`, `demos/adventure-client/`,
  `demos/adventure-scenario-test/`, `demos/adventure-chat-actors/`,
  `demos/adventure-npc-shopkeeper/`, `demos/adventure-npc-wanderer/`,
  `demos/remote-session-adventure-worker/` (only Adventure-side
  changes; the remote-session worker boundary is shared with the
  remote-session plan -- coordinate before changing the bridge), and
  any new `demos/adventure-*` crate
- Aurelian-specific Adventure interface extensions in `schema/capos.capnp`
  (shared serial surface -- see `docs/plans/README.md` Concurrency Notes;
  only one plan at a time may change `schema/capos.capnp`, and
  `make generated-code-check` must pass before the next plan picks it up)
- `system-adventure.cue` and the matching `make run-adventure` target
- `docs/demos/adventure.md`, `docs/backlog/aurelian-frontier.md`,
  `docs/proposals/aurelian-frontier-proposal.md`

Do not touch from this plan:

- `tools/remote-session-client/` (owned by remote-session plan)
- `demos/paperclips-*` (owned by Paperclips plan)
- Kernel-core cap tables, scheduler, or device manager
- Repo-root `system-*.cue` defaults migration scope (owned by
  system-config slice 3 plan), except `system-adventure.cue` direct edits

## Validation Commands

- `make fmt-check`
- `make generated-code-check`
- `cargo build --features qemu`
- `cargo test-config`
- `cargo test-lib`
- `cargo build-demos-capos`
- `make run-adventure`
- `make run-remote-session-adventure-interop` (required when the
  iteration changes `demos/remote-session-adventure-worker/` or the
  Adventure call shape consumed by the remote-session worker)

## Success Criteria

Each landed iteration must keep `make run-adventure` green, leave the
content tree deterministic, and update `docs/backlog/aurelian-frontier.md`
to mark the corresponding follow-up done with a commit hash. Game-depth
expansion past the deterministic mission slice stays opt-in until the user
explicitly redirects priority.

### Task 1: Pick the next backlog follow-up to land

- [ ] Read the open backlog items in `docs/backlog/aurelian-frontier.md`
      and pick the smallest one that still produces a visible scenario
      assertion (no architectural restructure unless the user asks).
- [ ] Sketch the proof shape: which `Adventure.*` calls change, which
      generated content fields move, and what the QEMU assertion checks.
      Capture the sketch as a commit-message-shaped paragraph in the task
      branch's worktree handoff.
- [ ] Confirm the slice does not require new kernel-side caps or a new
      schema interface; if it does, stop and route the prerequisite
      through `WORKPLAN.md`.

### Task 2: Implement the picked follow-up in adventure-content first

- [ ] Add or update generated content in
      `demos/adventure-content/content/prototype.cue` and the matching
      pure Rust validation in `demos/adventure-content/`.
- [ ] Cover the change with `demos/adventure-scenario-test/` (and any
      `demos/adventure-content/` host tests) for bounded transitions,
      replay safety, and any new admission helper.
- [ ] Keep all source-game identifiers, text, and exact balance values out
      of generated content; reuse the clean-room shape proven in earlier
      Aurelian phases.

### Task 3: Wire adventure-server and the scenario process

- [ ] Implement the matching server-side state mutation, idempotent
      replay, and stale-version rejection in `adventure-server`. Keep
      service-owned vs player-local state separated as in earlier phases.
- [ ] Update the real scenario process (the QEMU proof binary) so it
      drives the new `Adventure.*` calls and asserts the visible state in
      `Adventure.status`, `Adventure.inventory`, or the related view.
- [ ] Confirm `make run-adventure` passes, then snapshot the run-adventure
      log into the dedicated worktree.

### Task 4: Doc and changelog updates

- [ ] Mark the follow-up done in `docs/backlog/aurelian-frontier.md` with
      commit hash and minute-precision timestamp.
- [ ] If the follow-up changes the proof shape visible to operators,
      update `docs/demos/adventure.md`.
- [ ] Add a `docs/changelog.md` entry referencing the commit.

### Task 5: Optional second slice in the same iteration

- [ ] Only if the picked follow-up was small and the run-adventure log
      stayed bounded, pick a second open backlog item and repeat Tasks
      1-4 for it. Otherwise leave this checkbox unchecked and stop.
- [ ] Keep both slices in one task branch only when their changed file
      sets are disjoint and the proof binaries do not interleave state.